Constraints Based Convex Belief Propagation
Authors: Yaniv Tenzer, Alex Schwing, Kevin Gimpel, Tamir Hazan
NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments show that CBCBP outperforms the conventional consistency potential based approach, while being at least an order of magnitude faster. |
| Researcher Affiliation | Academia | Yaniv Tenzer Department of Statistics The Hebrew University Alexander Schwing Department of Electrical and Computer Engineering University of Illinois at Urbana-Champaign Kevin Gimpel Toyota Technological Institute at Chicago Tamir Hazan Faculty of Industrial Engineering and Management Technion Israel Institute of Technology |
| Pseudocode | Yes | Figure 1: The CBCBP algorithm. Shown are the update rules for the λ and ν messages. |
| Open Source Code | No | The paper discusses the use of a third-party open-source toolkit (Moses) but does not provide specific access to the authors' own implementation of CBCBP or experiment code. |
| Open Datasets | Yes | We evaluate our approach on the task of semantic segmentation using the MSRC-21 dataset [21] as well as the Pascal VOC 2012 [4] dataset. |
| Dataset Splits | No | The paper mentions using datasets for training and testing (e.g., WMT2011 for training, WMT2009 for testing) and refers to standard error measures, but it does not provide specific details on training/validation/test splits (e.g., exact percentages or sample counts for each split). |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions 'Moses' as a toolkit and implies implementation in 'Python' (given the context of ML papers and common tools), but it does not specify version numbers for any software dependencies or libraries. |
| Experiment Setup | Yes | We use maximum phrase length m = 3 and distortion limit d = 3. We run 250 iterations of CBCBP for each sentence. For the feature weights (γ), we use the default weights in Moses, since our features are analogous to theirs. |